Updated: 1 October 2026
AI-generated creative has moved from an experimentation question to a workflow question for European marketing teams.
Since 2 August 2026, Article 50 of the EU AI Act has applied transparency requirements to certain AI systems and AI-generated or manipulated content.
But the rule is easy to oversimplify.
It does not mean every image touched by an AI tool needs a giant “AI GENERATED” label.
The obligations depend on what was generated, who is providing or deploying the AI system, how realistic the content is, and what the audience is being shown.
For marketers, the useful response is not to stop using generative AI.
It is to put a disclosure check into the creative workflow before a campaign goes live.
First: know whether you are the provider or the deployer
Article 50 creates different responsibilities.
A company that develops or provides an AI system capable of generating synthetic text, audio, images or video may have provider obligations.
Among them is a requirement to make AI-generated or manipulated outputs detectable and marked in a machine-readable format, as far as technically feasible.
Most marketing teams using third-party AI tools are more likely to be acting as deployers: organisations using an AI system under their authority for professional purposes.
That distinction matters because the disclosure obligations marketers are most likely to encounter concern the way certain generated content is presented to people.
The European Commission has also published a voluntary Code of Practice designed to help providers and deployers demonstrate compliance with Article 50.
Run this five-question creative check
Before an AI-assisted asset goes live in the EU, ask these questions.
1. Does it depict something as real that did not happen?
Start with the highest-risk creative.
Article 50 specifically addresses AI-generated or manipulated image, audio or video that constitutes a deepfake.
The AI Act describes this as generated or manipulated content resembling existing people, objects, places, entities or events that would falsely appear authentic or truthful.
For marketers, that can reach beyond celebrity face swaps.
Examples worth escalating for review could include:
- a realistic person appearing to use or endorse a product when they did not;
- an AI-generated voice made to sound like a real individual;
- a realistic image of an actual location showing an event that never occurred;
- manipulated footage that makes a real event appear materially different.
If the asset falls into this territory, Article 50 says the deployer must disclose that it has been artificially generated or manipulated.
The disclosure must be clear and distinguishable and provided no later than the audience’s first exposure.
2. Is the creative obviously fictional or artistic?
Context matters.
The AI Act contains different treatment for content that forms part of an evidently artistic, creative, satirical, fictional or analogous work.
The transparency obligation does not simply disappear, but disclosure can be made in an appropriate way that does not interfere with the enjoyment or display of the work.
That means teams should avoid turning the process into a binary rule such as:
AI used = identical label on every creative.
Instead, record why the asset was classified and what disclosure method was selected.
The important operational improvement is documentation.
Someone reviewing the campaign later should be able to see what tool was used, what was generated, what was edited by people and why the team decided a particular disclosure treatment was appropriate.
3. Is AI-generated text informing the public about a matter of public interest?
Article 50 also covers certain AI-generated or manipulated text.
Where AI-generated text is published for the purpose of informing the public on matters of public interest, the deployer may have to disclose that the text was artificially generated or manipulated.
There is an important exception.
The AI Act says that disclosure requirement does not apply where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for the publication.
For a marketing or publishing team, that makes the editorial process relevant.
Do not treat “a human looked at it” as a magic compliance phrase.
Instead, establish who reviewed the material, who approved it and who holds editorial responsibility.
For NEMO itself, that is also why AI-assisted research should not replace editorial verification.
4. Are you preserving the information supplied by the AI system?
There is another layer marketers should not ignore.
Article 50 also requires providers of generative AI systems to make generated content machine-readable and detectable as artificially generated or manipulated, subject to technical feasibility and other considerations.
Marketing production pipelines can break that chain.
An image may pass through:
AI generator → Photoshop → design platform → compression → ad platform → publisher.
Metadata or technical signals may disappear along the way.
That does not mean a marketer automatically becomes responsible for the AI provider’s duties.
It does mean teams should stop treating provenance information as disposable by default.
If an AI tool supplies content credentials, metadata or another machine-readable indicator, determine whether your editing and export process preserves it.
This should become part of creative QA alongside dimensions, file size and tracking parameters.
The 15-minute campaign audit
For an existing campaign, open the creative folder and add these fields to the production sheet:
| Check | Record |
|---|---|
| AI system used | Tool/model |
| Asset type | Image / video / audio / text |
| AI contribution | Generated / manipulated / assisted |
| Real person, place or event depicted? | Yes / No |
| Could the asset reasonably appear authentic? | Yes / No |
| Public-interest text? | Yes / No |
| Human editorial review? | Yes / No |
| Editorial owner | Name/team |
| Disclosure required? | Yes / No / Review |
| Disclosure location | Creative / caption / landing page / other |
| Machine-readable marking preserved? | Yes / No / Unknown |
| Platform declaration completed? | Yes / No / N/A |
| Review date | Date |
Anything marked Unknown or Review becomes the queue.
That is far more useful than asking the creative team to memorise Article 50.
Do not confuse platform controls with the law
Google, Meta, TikTok, YouTube and other advertising or publishing platforms may have their own controls for declaring altered or synthetic content.
Those interfaces can help operationalise a campaign.
They are not the same thing as the AI Act itself.
A platform allowing an advertiser to tick an “AI generated” box does not by itself establish that every Article 50 obligation has been satisfied.
The reverse is also important: a platform’s disclosure policy can be broader than the legal requirement.
So maintain two checks:
Regulatory requirement
and
Platform requirement
Do not collapse them into one field.
What marketing teams should change now
The biggest change is procedural.
Add one AI-transparency gate between creative approval and campaign launch.
For each materially AI-generated asset:
- identify what the AI actually created or changed;
- determine whether real people, objects, places, entities or events are being represented;
- assess whether the content could falsely appear authentic;
- review public-interest text separately;
- document human editorial review;
- preserve available provenance or machine-readable information where appropriate;
- check the destination platform’s own synthetic-content controls;
- record the final disclosure decision.
This is not about putting a warning on every AI-assisted banner.
It is about making sure nobody discovers after launch that the team never asked the question.
What NEMO is tracking next
The regulation is only one part of the implementation story.
The next useful question for marketers is:
Where do the major advertising platforms actually put these disclosure controls?
NEMO will track how Google, Meta, TikTok and other major platforms expose AI-generated-content declarations to European advertisers, including where the setting appears and what type of creative triggers it.
That will turn the legal requirement into something marketers can check inside the tools they use every day.
Primary sources
European Commission — Guidelines on Article 50 transparency obligations
https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems
European Commission — Code of Practice on Transparency of AI-generated Content
https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content
European Commission — Article 50 Code of Practice FAQ
https://digital-strategy.ec.europa.eu/en/faqs/code-practice-transparency-ai-generated-content
EUR-Lex — Regulation (EU) 2024/1689, Article 50
https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689
This article provides operational information for marketing teams and is not legal advice.